fix: retry thinking-only empty outputs, centralize reference marker stripping

- ValidateTurn no longer errors on thinking-only responses, deferring to
  ShouldRetryEmptyOutput which now also covers thinking-only outputs.
- Empty output retry uses multi-turn follow-up with a regeneration prompt
  suffix and parent_message_id in the same DeepSeek session.
- Centralize StripReferenceMarkersEnabled into textclean package to
  eliminate duplicated hardcoded booleans across 4 protocol handlers.
- Log a deprecation warning when the legacy "compat" config key is used.
- Document thinking-only retry and reference marker stripping in API.md.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

internal/httpapi/claude/handler_routes.go

@@ -8,6 +8,7 @@ import (
 
 	"ds2api/internal/config"
 	dsprotocol "ds2api/internal/deepseek/protocol"
+	"ds2api/internal/textclean"
 	"ds2api/internal/util"
 )
 
@@ -22,7 +23,7 @@ type Handler struct {
 }
 
 func stripReferenceMarkersEnabled() bool {
-	return true
+	return textclean.StripReferenceMarkersEnabled()
 }
 
 var (

internal/httpapi/gemini/handler_routes.go

@@ -5,6 +5,7 @@ import (
 
 	"github.com/go-chi/chi/v5"
 
+	"ds2api/internal/textclean"
 	"ds2api/internal/util"
 )
 
@@ -19,7 +20,7 @@ type Handler struct {
 
 //nolint:unused // used by native Gemini stream/non-stream runtime helpers.
 func stripReferenceMarkersEnabled() bool {
-	return true
+	return textclean.StripReferenceMarkersEnabled()
 }
 
 func RegisterRoutes(r chi.Router, h *Handler) {

internal/httpapi/openai/chat/handler.go

@@ -12,6 +12,7 @@ import (
 	"ds2api/internal/httpapi/openai/history"
 	"ds2api/internal/httpapi/openai/shared"
 	"ds2api/internal/promptcompat"
+	"ds2api/internal/textclean"
 	"ds2api/internal/toolcall"
 	"ds2api/internal/toolstream"
 )
@@ -36,7 +37,7 @@ type streamLease struct {
 }
 
 func stripReferenceMarkersEnabled() bool {
-	return true
+	return textclean.StripReferenceMarkersEnabled()
 }
 
 func (h *Handler) applyCurrentInputFile(ctx context.Context, a *auth.RequestAuth, stdReq promptcompat.StandardRequest) (promptcompat.StandardRequest, error) {

internal/httpapi/openai/responses/handler.go

@@ -11,6 +11,7 @@ import (
 	"ds2api/internal/httpapi/openai/history"
 	"ds2api/internal/httpapi/openai/shared"
 	"ds2api/internal/promptcompat"
+	"ds2api/internal/textclean"
 	"ds2api/internal/toolstream"
 )
 
@@ -29,7 +30,7 @@ type Handler struct {
 }
 
 func stripReferenceMarkersEnabled() bool {
-	return true
+	return textclean.StripReferenceMarkersEnabled()
 }
 
 func (h *Handler) applyCurrentInputFile(ctx context.Context, a *auth.RequestAuth, stdReq promptcompat.StandardRequest) (promptcompat.StandardRequest, error) {